Final Project: Boston Weather Trend

Average Temperature

ggplot(data=boston, aes(x=YEAR,y=month_ordered)) + 
  geom_tile(aes(fill = TAVG),colour = "white") + 
  scale_fill_gradientn(colours=rev(brewer.pal(10,'Spectral')), na.value = "grey98",
                       limits = c(-20, 100)) + 
  theme(legend.title=element_blank(), axis.title.y=element_blank(), axis.title.x=element_blank(), plot.title = element_text(hjust = .5)) + 
  ggtitle("Average Temperature/˚F of Boston from 1936-1-1 to 2018-5-1 (Based on Daily Average)")+
  scale_x_continuous(breaks=seq(1936,2018,5))

Max Temperature

ggplot(data=boston, aes(x=YEAR,y=month_ordered)) + 
  geom_tile(aes(fill = TMAX),colour = "white") + 
  scale_fill_gradientn(colours=rev(brewer.pal(10,'Spectral')), na.value = "grey98",
                       limits = c(-20, 100)) + 
  theme(legend.title=element_blank(),axis.title.y=element_blank(),axis.title.x=element_blank(),plot.title = element_text(hjust = .5)) + 
  ggtitle("Maximum Temperature/˚F of Boston from 1936-1-1 to 2018-5-1 (Based on Daily Max)")+
  scale_x_continuous(breaks=seq(1936,2018,5))

Min Temperature

ggplot(data=boston, aes(x=YEAR,y=month_ordered)) + 
  geom_tile(aes(fill = TMIN),colour = "white") + 
  scale_fill_gradientn(colours=rev(brewer.pal(10,'Spectral')), na.value = "grey98",
                       limits = c(-20, 100)) + 
  theme(legend.title=element_blank(),axis.title.y=element_blank(),axis.title.x=element_blank(),plot.title = element_text(hjust = .5)) + 
  ggtitle("Minimum Temperature/˚F of Boston from 1936-1-1 to 2018-5-1 (Based on Daily Min)")+
  scale_x_continuous(breaks=seq(1936,2018,5))

According to the temperature gradient maps above, temperature in June, July and August is obviously higher than other months, while December, January and February generally have the lowest temperature. December, 1962 stands out for its extradinarily low temperature.

Precipitation

ggplot(data=boston, aes(x=YEAR,y=month_ordered)) + 
  geom_tile(aes(fill = PRCP),colour = "white") + 
  scale_fill_gradientn(colours=brewer.pal(9,'BrBG'), na.value = "grey50",
                       limits = c(0, 7)) + 
  theme(legend.title=element_blank(),axis.title.y=element_blank(),axis.title.x=element_blank(),plot.title = element_text(hjust = .5)) + 
  ggtitle("Pricipitation/inch of Boston from 1936-1-1 to 2018-5-1 (Based on Daily PRCP)")+
  scale_x_continuous(breaks=seq(1936,2018,5))

As shown above, the overall precipitation in Boston is not too much. There are some months standing out with high precipitation, among which May, 1984 is the top 1 month with highest precipitation.

Average Temperature

myts <- ts(monthly$TAVG,start=c(1936,1), end=c(2018,4), frequency=12)
ts.decompose <- decompose(myts)
autoplot(ts.decompose)+ggtitle("Time Sereis of Average Tempetation")

## Warning: attributes are not identical across measure variables;
## they will be dropped

## Warning: Removed 24 rows containing missing values (geom_path).

dygraph(ts.decompose$x) %>% dyRangeSelector() %>% dyOptions(drawGrid = F)

dygraph(ts.decompose$trend) %>% dyRangeSelector() %>% dyOptions(drawGrid = F)

dygraph(ts.decompose$seasonal) %>% dyRangeSelector() %>% dyOptions(drawGrid = F)

dygraph(ts.decompose$random) %>% dyRangeSelector() %>% dyOptions(drawGrid = F)

Precipitation

ts_prcp <- ts(na.omit(monthly$PRCP),start=c(1936,1), end=c(2018,4), frequency=12)
autoplot(decompose(ts_prcp))+ggtitle("Time Sereis of Precipitation")

## Warning: attributes are not identical across measure variables;
## they will be dropped

## Warning: Removed 24 rows containing missing values (geom_path).

The time series of precipitation demonstrates a less seasonal pattern. But it has several peaks with an interval of 30 years. For instance, the year of 1955 or os has a peak, and around 30 years later, precipitation reached another peak in 1983, which is followed by the next peak in 2010.

Correlation between Temperature and Precipitation

ggplot(data=monthly,aes(x=TAVG,y=PRCP,color=factor(month_ordered))) + 
  geom_point(alpha=.5) + scale_color_manual(values=mycols) + 
  xlab('Temperature') + ylab('Precipitation')

## Warning: Removed 1 rows containing missing values (geom_point).

ggplot(data = monthly,mapping = aes(x=TAVG,y=PRCP)) + geom_point()+
  geom_smooth(method = "glm", method.args = list(family = "gaussian"))+ 
  xlab('Temperature') + ylab('Precipitation')

## Warning: Removed 1 rows containing non-finite values (stat_smooth).

## Warning: Removed 1 rows containing missing values (geom_point).

cor.test(monthly$TAVG, monthly$PRCP, use = "complete.obs")

## 
##  Pearson's product-moment correlation
## 
## data:  monthly$TAVG and monthly$PRCP
## t = -3.8407, df = 986, p-value = 0.0001305
## alternative hypothesis: true correlation is not equal to 0
## 95 percent confidence interval:
##  -0.18239623 -0.05949065
## sample estimates:
##        cor 
## -0.1214088

From the above plots, we can come to the conclusion that the temperature is barely correlated to the precipitation of Boston. I also conduct correlation test on these two variables, the result of which shows that the r value is -0.121, a very weak negative relationship between these two variables. The precipitation among the 12 months has no obvious difference.

Correlation between Wind Speed and Temperature

ggplot(data=monthly,aes(x=AWND,y=TAVG,color=factor(month_ordered))) + 
  geom_point(alpha=.5) + scale_color_manual(values=mycols) + 
  xlab('Wind speed') + ylab('Temprature')

## Warning: Removed 576 rows containing missing values (geom_point).

ggplot(data = monthly,mapping = aes(x = AWND, y = TAVG)) + geom_point()+
  geom_smooth(method = "glm", method.args = list(family = "gaussian")) + 
  xlab('Wind speed') + ylab('Temprature')

## Warning: Removed 576 rows containing non-finite values (stat_smooth).

## Warning: Removed 576 rows containing missing values (geom_point).

cor.test(monthly$TAVG, monthly$AWND, use = "complete.obs")

## 
##  Pearson's product-moment correlation
## 
## data:  monthly$TAVG and monthly$AWND
## t = -16.233, df = 411, p-value < 2.2e-16
## alternative hypothesis: true correlation is not equal to 0
## 95 percent confidence interval:
##  -0.6804793 -0.5624534
## sample estimates:
##        cor 
## -0.6250255

The above plots demonstrate that there is a strongly negative correlation between the temperature and wind speed in Boston. As the wind speed increases, the average temperature decreases. Correlation test is conducted on these two variables. the result shows that the r value is -0.625, a strong negative relationship between these two variables. Besides, the second plot shows an obvious difference in terms of the wind speed among different months. High wind speed are likely to concentrate in January and February, while June, July and August have relatively low wind speed.

Correlation between Snow and Temperature

ggplot(data=monthly,aes(x=TAVG,y=SNOW,color=factor(month_ordered))) + 
  geom_point(alpha=.5) + scale_color_manual(values=mycols) + 
  xlab('Temprature') + ylab('Snow')

ggplot(data = monthly,mapping = aes(x = TAVG, y = SNOW)) + geom_point()+
  geom_smooth(method = "glm", method.args = list(family = "gaussian"))+ 
  xlab('Temprature') + ylab('Snow')

cor.test(monthly$TAVG, monthly$SNOW, use = "complete.obs")

## 
##  Pearson's product-moment correlation
## 
## data:  monthly$TAVG and monthly$SNOW
## t = -6.9185, df = 987, p-value = 8.2e-12
## alternative hypothesis: true correlation is not equal to 0
## 95 percent confidence interval:
##  -0.2737319 -0.1548030
## sample estimates:
##        cor 
## -0.2150647

From the above plots, we can come to the conclusion that the temperature is slightly negatively correlated to the amount of snowfall in Boston. As the temperature goes up, the amount of snowfall decreases. I also conduct correlation test on these two variables, the result of which shows that the r value is -0.215, a weak negative relationship between these two variables. Besides, the second plot shows that the snowfall in Boston concentrates in three months: December, January and February.